Abstract:

In traditional well testing, pressure or flow rate transient data are collected for a short duration. Unlike traditional well testing tools, modern permanent downhole gauges have the capability to provide continuous measurements of both pressure and flow rate for several years or longer. Thus, data from these tools are a rich source of information for well testing interpretation. However, the large volume of information also brings with it a large volume of noise. Therefore, there is the need for the analysis of a combined pressure and flow rate data model for denoising usage. Usually, reservoir engineering models start from the defining physical equations, deduce a relationship between the flow rate and pressure, and then treat remaining deviations in the data as noise. By this approach, bias can enter the data by the imposition of inappropriate physics. As an alternative approach, this study tried to start from the data, either with the aid of wavelet or data mining techniques, to construct an empirical model for the flow rate and pressure. The study investigated whether the empirical model of data pairs may provide a better way of characterizing the true reservoir response.

This study first investigated wavelet methods in the identification of noisy data, and discussed the advantages and limitations of its performance in the denoising process. Then the study looked into the application of data mining techniques in constructing the empirical model. The study addressed the application of data mining both in the real time space and in the Laplace space. The data mining in the Laplace space provides an entirely new way of examining and understanding well test measurements. The Laplace space approach successfully extracted the correct pressure transient from noisy signals.

Press the Back button in your browser.

Copyright 2009, Yang Liu: Please note that the reports and theses are copyright to their original authors. Authors have given written permission for their work to be made available here. Readers who download reports from this site should honor the copyright of the original authors and may not copy or distribute the work further without the permission of the author, Yang Liu.